Researchers at Shanghai Jiao Tong University have achieved a 29.3kbps key rate using continuous-variable quantum key distribution (CV-QKD) over a 100-kilometer fiber link. This fully integrated system, incorporating silicon transmitters and receivers, establishes new benchmarks for both distance and key rate in chip-based quantum communication. The work unifies on-chip integration, room-temperature operation and provable security, addressing key bottlenecks in long-distance QKD. By demonstrating these capabilities, the team advances the development of practical, scalable quantum networks.
Integrated CV-QKD System Achieves 29.3kbps Key Rate at 100km
A composable secret-key rate of 29.3kbps was certified at a 100km distance, a new benchmark for integrated continuous-variable quantum key distribution (CV-QKD) systems, achieved through a fully integrated platform incorporating on-chip lasers, transmitters and receivers. This demonstration moves beyond prior CV-QKD implementations that often relied on discrete, bulky components. The system’s performance was validated through a 140-billion-symbol block, rigorously incorporating measured channel statistics and calibrated receiver parameters into a composable finite-size analysis.
The achieved key rate stems from a system designed for robust phase lock, maintained throughout the long data records necessary for composable security, and consistently outperformed conventional unscented Kalman filters. A Bayesian machine-learning algorithm was central to this stability, enabling the system to approach the Shannon limit using rate-matched multidimensional reconciliation.
At 125km, the system still generated a 12.9kbps key rate under finite-size analysis, while at 150km, the rate reached 9.17kbps under asymptotic analysis, collectively establishing the longest reported reaches for integrated CV-QKD in these security regimes. The reconciliation stage itself assigns each codeword to a decoding route based on publicly disclosed signal-to-noise ratio, optimizing performance.
This integrated platform’s success is a validated operating point where the complete key-distillation chain was executed and verified. By demonstrating a composable secret key directly from data acquired by this integrated platform, the work represents a step towards realizing practical, chip-based quantum networks capable of secure long-distance communication.
Hybrid III-V/Silicon Nitride Integration for CV-QKD Terminals
Hybrid III-V/silicon nitride (Si₃N₄) integrated lasers underpin the system’s performance, a design choice that allowed for the creation of a fully integrated terminal capable of operating over extended fibre links. These lasers, coupled with silicon transmitters and receivers, facilitated operation with a local local oscillator (LLO) across distances ranging from 25 to 150 kilometers, a significant expansion of practical range for integrated CV-QKD systems.
The system implements a fully integrated CV-QKD platform in which two hybrid III–V/Si3N4 integrated lasers, a silicon transmitter, and a silicon coherent receiver implement the core terminal functions, operating with a local local oscillator (LLO) over fibre links of 25–150 km.
The design of the hybrid lasers addresses a critical challenge in long-distance CV-QKD: maintaining ultra-low noise levels. Each laser combines a commercial distributed feedback (DFB) gain chip with a low-loss silicon-nitride feedback circuit incorporating a high-Q microring resonator. This resonator, featuring a 3-μm-wide waveguide, expands the mode field to reduce optical intensity at interfaces and minimize propagation loss, while Euler bends smooth transitions between waveguide sections to suppress higher-order modes within a compact footprint.
Local Oscillator Configuration for Enhanced CV-QKD Security
While integrated transmitters, receivers, and high-symbol-rate links have been demonstrated individually, previous 100-kilometer LLO operation relied on bulk, discrete components rather than full on-chip integration. This work distinguishes itself by achieving composable finite-size security in a chip-based CV-QKD system, a feat previously limited to shorter transmission distances. Beyond this benchmark, the platform sustains key rates of 12.9 kbps at 125 kilometers under the same finite-size security regime, and 9.17 kbps at 150 kilometers when analyzed under asymptotic conditions.
At the receiver, a second integrated laser generates the local oscillator, with an optical switch alternating between quantum and vacuum windows to acquire shot-noise traces. Conventional forward-only phase recovery methods estimate based on past samples, neglecting subsequent pilot observations; this approach can introduce errors that directly contribute to excess noise.
The team’s system addresses this limitation by exploiting all available data, improving phase estimation, particularly at low signal-to-noise ratios over long links. These advancements, combined with the integrated platform, establish a foundation for scalable and secure quantum communication networks.
Bayesian Machine Learning for Robust Phase Lock Recovery
A Bayesian machine-learning algorithm proved critical in sustaining a secure quantum connection over 100 kilometers, consistently exceeding the performance of traditional phase-tracking methods. This advancement addresses a core challenge in continuous-variable quantum key distribution (CV-QKD): maintaining accurate phase lock despite noise and signal degradation over extended distances, a necessity for composable security. 29.3 kbps at 100 km from a 140-billion-symbol block, with 12.9 kbps at 125 km under finite-size analysis and 9.17 kbps at 150 km under asymptotic analysis.
Maintaining phase stability is particularly difficult in a local oscillator (LLO) configuration, where the receiver must estimate the relative phase between two free-running chip lasers. Over long fiber links, even slight errors in this estimation can disrupt key generation before it begins. The team’s approach uses a machine-learning-based algorithm that tracks phase across extended data records, essential for the rigorous security demands of composable QKD.
This algorithm demonstrably lowers phase-estimation error compared to the conventional unscented Kalman filter, a standard technique in signal processing. Rate-matched multidimensional reconciliation, operating near the Shannon limit, further optimizes data transmission and minimizes information loss. The system’s success relies on more than just algorithmic improvements; it is a complete integration of hardware and software. Ten-Hz-level integrated lasers minimize relative phase diffusion, keeping the phase within a trackable range for the state-space model.
This, combined with a stabilized transmitter and calibrated balanced detection, reduces noise and enhances signal clarity. The entire digital chain, from phase recovery through reconciliation to privacy amplification, is co-designed within the integrated system, ensuring smooth operation and optimized performance. “Extracting a valid secret key therefore requires combining device-level noise suppression with algorithmic recovery to ensure composable security across a continuous data record,” the researchers state.
The system employs a silicon-on-insulator (SOI) photonic integrated circuit co-packaged with high-speed electronics within a system-in-package (SiP) platform. One laser at Alice is the coherent-state carrier, while the other at Bob functions as the local oscillator, eliminating the need for a bright phase reference transmitted through the fiber. This configuration, coupled with the robust phase-lock recovery, establishes a foundation for scalable and secure chip-based quantum networks.
Composable Security Analysis & Finite-Size Key Rate Performance
The platform also achieved 12. The composable security analysis incorporates several factors for rigorous evaluation, including decoding failures and information revealed during reconciliation, detailed in Supplementary Note 2. A Bayesian machine-learning algorithm maintains robust phase lock throughout the long records required for composable security, consistently outperforming the conventional unscented Kalman filter, while rate-matched multidimensional reconciliation approaches the Shannon limit.
Long-Distance CV-QKD Challenges: Noise Margin & Phase Estimation
Maintaining a stable phase lock presents a significant hurdle in long-distance continuous-variable quantum key distribution (CV-QKD), particularly as signal strength diminishes with fibre length. A slight error in pilot signal estimation can disrupt this lock before key extraction, and sustaining it throughout the extended data records needed for robust, composable security is especially challenging. The work demonstrates that a Bayesian machine-learning algorithm consistently outperforms conventional unscented Kalman filters in maintaining this critical phase lock across long records, a necessity for secure communication.
This algorithm’s effectiveness stems from its ability to adapt to the subtle changes in the system over time, ensuring reliable performance even at the low signal-to-noise ratios inherent in 100-kilometer transmissions. Successfully extracting a secure key at these distances requires a combined approach of suppressing noise at the device level and employing algorithms capable of recovering information from noisy signals.
The demonstrated system achieves this by co-designing integrated lasers, transmitters, and receivers with a complete digital chain, encompassing phase recovery, reconciliation and privacy amplification. Specifically, the integrated lasers, hybrid III-V/Si₃N₄ self-injection-locking (SIL) lasers, were designed for the ultra-low noise levels demanded by long-distance CV-QKD. The system’s performance is further enhanced by design choices that address propagation losses and maintain signal integrity.
At 100 kilometers, a block of 1.4× 10¹¹ symbols, acquired at = 6.77 SNU, was used for parameter estimation in the composable finite-size analysis. The ability to achieve a positive finite-size composable key rate at this distance demonstrates the effectiveness of the integrated system and the sophisticated algorithms employed.
👉 More information
🗞 Fully integrated continuous-variable quantum key distribution with composable security over 100 km
✍️ Yankai Xu, Xinhang Li, Tao Wang, Yuyao Guo, Peng Huang, Linjie Zhou and Guihua Zeng (State Key Laboratory of Photonics and Communications, Shanghai Jiao Tong University, Shanghai, China); Jisheng Dai and Xueqin Jiang (College of Information Science and Technology, Donghua University, Shanghai, China)
🧠 ArXiv: https://arxiv.org/pdf/2608.29200




See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
